Griffith University
Griffith University
Griffith University
Project Title:
“Numerical study of flow over a smooth circular cylinder
and a square cylinder at low-high Reynolds number”
Student Name:
Abstract
The drag effect is a broad term used to control flow and evaluate the economic
design of aerodynamics, automobiles, chimneys, buildings, and hydraulic structures,
among other things. Two cases are considered for the problem of steady,
incompressible two-dimensional flow: There are two types of cylinders: cylindrical and
square. The flow around the smooth cylinder and square cylinder with a characteristic
length of 25 mm, placed horizontally, perpendicular to the main flow direction, was
numerically analysed and quantified at various Reynolds numbers based on the
characteristic length and calculated free stream velocity in the range of 1<Re<1X105.
For solving the 2D governing equation for the tested model, the numerical analysis
obtained by CFD using ANSYS FLUNET with K-omega turbulence model was
considered. The current study investigated that as the Re of the flow increases,
different instability regimes are observed for both cases. The prediction of drag
coefficient, velocity contour, and pressure contour shows good agreement with
previous published numerical investigations.
1.0 Introduction
Flow separation over a bluff body is a classic fluid mechanics problem that predicts
drag and lift coefficients, wake formation, and pressure fields [1-5]. Flow around a
circular cylinder and square is critical for engineers and researchers. Furthermore,
flow over circular cylinders can be modelled in a variety of applications such as heat
exchangers, cooling towers, nuclear cooling systems, bridge piers, and subsea
pipelines [6-9]. The majority of engineering problems have a sub-critical Reynolds
number. The shedding of vortices causes boundary layer separation and flow
oscillation in the wake zone behind the cylinder at moderate and high Reynolds
numbers [10-12].
To better understand the complexities of laminar and turbulent flows around circular
cylinder and square cylinder, new techniques and definitions have been developed.
Computational fluid dynamics (CFD) is a useful tool for displaying the flow field in terms
of streamlines, contour lines, velocity vectors, and velocity profiles for various flow
conditions. CFD techniques can also be used to calculate the drag coefficient,
pressure coefficient, and forces acting on the structures. The numerical analyses in
this study were carried out using the k-omega turbulence closure model and
commercial CFD software.
The figures 2a and 2b represent the computational domain for this study. The domain
in this study is circular and square, with a length of 50D units and a width of 10D units,
where D stands for characteristic length, and the value of D for circular cylinder
diameter and square cylinder side is 25mm. Inside the rectangle, the cylinder and
square are placed, with the centre at (15D, 5D). The flow is from left to right, with the
cylinder and square positioned at the domain's centreline.
4.2 Boundary Conditions
The figures 3a and 3b show the physical boundary conditions for the problem
configuration for cylinder and square sections, respectively.
Figure 3: Boundary conditions for the fluid domain :(a) Cylinder, (b) Square
The computational domain's left wall (A) is assigned as the inlet. At the inlet boundary
with free stream velocity, i.e., V, the velocity inlet boundary condition is assigned. The
domain's walls (B), cylinder (D), and square (D) have no slip boundary conditions for
flow at their surfaces, as well as the domain's upper and lower surfaces, i.e., u=0, v=0.
The domain's extreme right surface is designated as an outlet. The pressure-outlet
boundary condition of the Dirichlet type pressure boundary condition (p = 0) is used at
the exit boundary with the zero input value of the static gauge pressure, i.e., u/x=v/y=0.
4.3 Discretization
For accurately predicting flow physics and ensuring the numerical solution's
stability, the computational mesh used to discretize the geometry is critical. Mesh
discretization is classified into two types: structural and unstructured mesh.
Structured two-dimensional mesh: It features a quadrilateral mesh structure and
provides simplicity and efficiency while requiring less memory. The major drawback is
that it is difficult to compute a structured mesh for a complicated geometric domain,
and the number of components must be increased to produce more accurate results.
The figure 4, shows an O grid [33] is specified for the cylinder and square cylinder
zone. The unstructured two-dimensional mesh type, with a triangular mesh structure,
is commonly used for complex designs. Figure 5, shows the unstructured mesh with
a 4mm element size for both geometries.
In the current study, two cases with each design were simulated to achieve a fine
resolution of the region. The basic output mesh information for both structured and
unstructured mesh is shown in table 1.
The deviation skewness and orthogonality matrices are used to evaluate mesh
deviation. The orthogonal quality is close to zero and better cells have close to one
because the skewness value is low, indicating higher element quality and poor cells.
Because the skewness and orthogonal quality in table-1 vary, the result may have a
different range.
4.4 Properties of the fluid
In this work, the fluid is considered as air and its properties are density=1.164 kg/m3,
and viscosity is 1.872X10-5 kg/m-s.
4.5Assumptions
The basic assumptions used to derive the theoretical equations are
summarized as follows:
- The fluid is considered as Newtonian fluid
- steady state, incompressible flow
- two dimensional flow/geometry
- Isothermal condition
- No body forces or gravitational effect
4.6 Governing equations
The flow past a circular and square has been simulated by solving numerically the
navier stokes equations for incompressible fluid in 2-D geometry. The equations for
continuity and momentum is expressed as follows:
o Continuity: + =0;
o Momentum:
- X-momentum, +𝑢 +𝑣 =− +𝑣 +
- Y-momentum, +𝑢 +𝑣 =− +𝑣 +
The drag coefficient (Cd ) and lift coefficient (Cl )is given by
𝐹 𝐹
𝐶 = 𝑎𝑛𝑑 𝐶 =
1 1
2 × 𝜌 × 𝑉 × 𝐷 2×𝜌×𝑉 ×𝐷
Where Fd = drag force, Fl=Lift force, D=Characteristics length, V=velocity in m/s,
ρ=Density in kg/m3. The fluid properties like density, viscosity of the streaming are
dependent on the Reynolds number and it is Defined as Inertia force to the viscous
force
𝜌𝑉𝐷
𝑅𝑒 =
𝜇
Where, Re=Reynolds Number, ρ=Density in kg/m3, V=velocity in m/s, D=
Characteristic length mm or m, μ=Viscosity of fluid Pa-s
4.7 Turbulence model: k-ω model
When compared to the k-ε model [28,32], the model solves the transport
variable such as k turbulence kinetic energy, and ω, turbulence dissipation rate, giving
more accurate results near the wall treatment under low-high Reynolds number.
𝜕(𝜌𝑘) 𝜕 𝜌𝑢 𝑘 𝜕 𝜌𝑘 𝜕𝑘 𝜕𝑢
+ = 𝜌𝑃 − 𝛽 ∗ 𝜌𝜔𝑘 + 𝜇+𝜎 , 𝑤𝑖𝑡ℎ 𝑃 = 𝜏 ,
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜔 𝜕𝑥 𝜕𝑥
𝜕(𝜌𝜔) 𝜕 𝜌𝑢 𝜔 𝛼𝜔 𝜕 𝜌𝑘 𝜕𝜔 𝜌𝜎 𝜕𝑘 𝜕𝜔
+ = 𝑃 − 𝛽𝜌𝜔 + 𝜇+𝜎 +
𝜕𝑡 𝜕𝑥 𝑘 𝜕𝑥 𝜔 𝜕𝑥 𝜔 𝜕𝑥 𝜕𝑥